Statistical analysis of feature-based molecular networking results from non-targeted metabolomics data

Area of Science:

• Metabolomics
• Computational Biology
• Bioinformatics

Background:

• Feature-based molecular networking (FBMN) is widely used for analyzing untargeted metabolomics data from liquid chromatography-tandem mass spectrometry (LC-MS/MS).
• Downstream statistical analysis of FBMN data presents a significant bottleneck, particularly for researchers new to statistical methods.
• Handling and interpreting complex data matrices generated by FBMN requires specialized guidance.

Purpose of the Study:

• To provide a comprehensive guide for the statistical analysis of FBMN results.
• To offer practical solutions for data handling, cleanup, normalization, and statistical interrogation of FBMN output tables.
• To lower the barrier to entry for new users in metabolomics data analysis.

Main Methods:

• Detailed explanations of data structure, cleanup, and normalization principles for FBMN data.
• Uni- and multivariate statistical analyses applied to FBMN results.
• Provision of code in R, Python, and QIIME2, shared via Jupyter Notebooks.

Main Results:

• A complete protocol, code repository (Jupyter Notebooks), and a web application with a graphical user interface are provided.
• The resources facilitate seamless integration, cleanup, and advanced statistical analysis of FBMN data.
• Demonstration using a previously published environmental metabolomics dataset.

Conclusions:

• The developed toolbox empowers new users to effectively analyze FBMN data and uncover molecular insights.
• The protocol is adaptable for various mass spectrometry-based metabolomics workflows beyond Global Natural Products Social Molecular Networking.
• Enhanced accessibility to advanced statistical analysis for untargeted metabolomics data.